Polyethylene fix deform question.

Dear Lammps Users,

Please help if you have some experience with the fix deform commands. Generally speaking, The structure is not deforming correspond to the simulation box. I used keyword “remap x”. So my question is with this fix deform command, It there anything that I need to beware of? Do I need to use a tight cube box? If you are interested, please email me back and I will send you the .lammpstrj dump file which you could check in VMD to understand what I am saying. They are less then 20MB.

For detailed information, please read the following:
I followed this tutorial which is about amorphous Polyethylene deformation:
https://icme.hpc.msstate.edu/mediawiki/index.php/Deformation_of_Amorphous_Polyethylene

here is the input script website:https://icme.hpc.msstate.edu/mediawiki/index.php/LAMMPS_Input_Deck_for_Atomistic_Deformation_of_Amorphous_Polyethylene

Basically, it is first doing 4 stages of equilibration with NVE, NPT, NPT and NPT. secondly fix NPT at a specific temperature and deform the simulation box in x direction, then measure the pressure on the boundary and generate the stress-strain curve.

By changing the running steps in each Equilibration stage to ten times larger (e.g. 10000 => 100000 running steps). By also using their input data file (it is 1000 atoms/chain *10 chain, totally 10000 atoms ) I successfully repeated the researcher’s result.

Now I generated a polyethylene structure data file (also 10000 atoms with 10 chain) with Avogadro and VMD, then run in Lammps with the exactly same input script, but sometimes when I checked the result, I found the PE structure only deformed a little bit, the deformation is not correspond to the simulation box.

I have tried changing the initial box as a cube or a rectangle before doing the equilibration. Which the cube box is kind of loose, you have 10 A between the box boundary and the nearest inner atoms in x direction and tight in y,z direction. The rectangle simulation is a tight fit of the atom structure. which each direction has maximum 1-2 A, and the length of box on y and z direction are equal.

The result is, for the cube box, I got only like 0.03 strain. For the rectangle box, I can got 0.2 strain, then the box deform but the PE structure did not move very much. then I tried both case with deformation in the Y axis and in all cases they deformed very well.

Thanks to all of you.

Yueqi Hu

Hi,

Usually fix deform command its a good choice for Periodic systems or in the case your simulating The effect of a shock loading (NVT) e.g Quasi-isentropic compression

Since your trying to deform a Polythylene structure I would recommend the following steps :

1). Do not use a periodic box at all, and definitely not in the length direction You want free surfaces to mimic the surface effects.

2). Do not use an NPT ensemble - this constrains the pressure (and therefore the stress) of your structure, which is also incorrect. Typically you would first do an NVT stage where you equilibrate the nanowire at a certain temperature (say 300K), after which you would do the elongation in an NVE ensemble. Neither of these imposes any constraints on the pressure/stress.

3). I would deform the polythylene structure by applying a ramped velocity

A Salute
Oscar G